The Leontief Open Economy Production Model

We will look at the idea for Leontief's model of an “open economy”  a term which we
will explain below. It starts by looking very similar to an example of a “closed
economy” that we saw near the beginning of this course. Here, however, we start with a
matrix whose columns represent “consumption” (rather than “production” as in the
earlier example).

In the case of simple examples, we can solve the equations that we will set up. However,
proofs of the deeper and more interesting theorems about “what happens in general” are
beyond the scope of this course.

Assume that an economy has 8 sectors W" ß ÞÞÞß W8 . Amounts consumed or produced by
each sector are measured in $ (value of goods consmbed or produced).

We begin with a consumption matrix G : the 4>2 column lists the amount of goods
(in $) consumed Ðor “demanded”Ñ by sector W4 from sectors W" ß W# ß ÞÞÞß W8 to produce one
unit ($) of its product..
Columns of G are
consumption vectors by sectors
W" â W4 â W8
to produce 1 unit of product (measured in $)


Æ Æ Æ
 W" Ä  "" "8 

ã  
- â - "4 â -
 
Consumed from  W3 Ä  -3" â -34 â -38  œ G
ã

  
ã
 W8 Ä  -8" â -84 â -88 
ã

Of course, each -34   !Þ

 -"" 
 ã 
 
For example, Column " œ  -3"  lists the consumption (demands) of Sector 1 from
 
 -8" 
ã

Sectors W" ,..., W8 to produce one unit ($) of product. Of course, we could also think of
this as a list of costs ($) for Sector 1 to produce one unit of product, and the sum of the
entries gives the total cost of all products consumed (demanded) by Sector 1 to produce
one unit of its product.
The same is true for each column: the entries in Column 4 represent the consumption
(demands, or costs) of sector W4 from sectors W" ß W# ß ÞÞÞß W8 in order for W4 to produce one
unit ($) of product.

Each column sum represents the costs for that sector to produce one unit. If a column
sum is  ", we say that sector is profitable.)
 -"4   B4 -"4 
 ã   ã 
   
If we weight Column 4 with a scalar B4   0 , then the vector B4  -34  œ  B4 -34  lists
   
 -84   B4 -84 
ã ã

the consumption (demand, cost) by sector W4 from other sectors to produce B4 units of its
own product.

Suppose B" ß B# ß ÞÞÞß B8 (   ! ) are amounts to be produced by sectors W" ß W# ß ÞÞÞß W8 . We

 B" 
B 
call B œ  #  is called a production vector. Then what does the vector G B mean?
 B8 
ã

It is a linear combination of the columns of G with the B3 's as weights:

 -""   -"#   -"8   -"" B"  ÞÞÞ  -"8 B8 

 ã   ã   ã   
       
G B œ B"  -31   B#  -3#   ÞÞÞ  B8  -38  œ  -3" B"  ÞÞÞ  -38 B8 
ã
       
 -81   -8#   -88   -8" B"  ÞÞÞ  -88 B8 
ã ã ã ã

 total demanded from sector W" by sectors W" ß ÞÞÞ ß W to produce B 

 
 
œ  total demanded from sector W# by sectors W" ß ÞÞÞ ß W8 to produce B 
ã
 
 total demanded from sector W8 ,C sectors W" ß ÞÞÞ ß W8 to produce B 
ã

In other words, GB lists the demands that must be supplied from each sector to the others
to deliver the production vector B Þ

The equation G B œ B asks for an “equilibrium value”  an production vector B for

which everything is “in balance,” where the list B of products (in $) from each sector
œ the list G B of total demands ($) from the sectors. For equilibrium, the production
from each sector must be not too large, not too small, but “just right.” This is a closed
economy  everything is just exchanged among the sectors.
Now we add another feature to the economic model: an open sector. It is nonproductive,
that is, it produces nothing that sectors W" ß ÞÞÞß W8 use. The open sector simply “demands”
the prodcuts of sectors W" ß ÞÞÞß W8 Þ (For example, these could be demands of the
government, demands for foreign export, for individual consumption, for charitable
groups, ...)

 ." 
. 
Supppose . œ  #  tells how many units (in $) the open sector demands from each
 .8 
ã

sector Wß ÞÞÞß W8 Þ The vector . is called the final demand vector

We want to know if it is possible to set a level of production B so that both the productive
and open sectors are satisfied, with nothing left over. That is, we want to find an B so that

Leontief Open Economy Production Model

B œ GB  . (***)
Å Å Å
Total production œ Demand from + Demand from
productive sectors open sector
to produce B
(called “intermediate ( called “final
demand” ) demand” )

If M is the identity matrix, we can rewrite this equation as

: MB œ GB  .
M B  G B œ . or

ÐM  GÑ B œ .

To take a simple example, suppose the economy has 4 productive sectors W" ß ÞÞÞW% and
the consumption matrix is
Consumption vectors for
W " W # W $ W%
Æ Æ Æ Æ
 Þ"! Þ!& Þ$! Þ#! 
 Þ"& Þ#& Þ!& Þ"! 
G œ 
 Þ"& Þ#! Þ"! Þ#! 
Þ$! Þ"! Þ"! Þ#&

ÐThe column sums in G are all  " , so each sector is profitable.)

  Þ*!  Þ#! 
  Þ"&  Þ"! 
 Þ!&  Þ$!
ÐM  GÑ œ  
Þ(&  Þ!&
Then
  Þ"& Þ)! 
 Þ$!  Þ"! Þ*!  Þ#&
 Þ#!  Þ"!

 #&!!! 
 "!!!! 
If the demand from the open sector is . œ  , then we want to solve
 &!!!! 
$!!!!

 Þ*!  Þ#!   #&!!! 

  Þ"&  Þ"!   "!!!! 
 Þ!&  Þ$!
ÐM  GÑB œ  B œ  œ.
Þ(&  Þ!&

  Þ"& Þ)!   &!!!! 

 Þ$!  Þ"! Þ*!  Þ#& $!!!!
 Þ#!  Þ"!

Row reducing the augmented matrix gives

 Þ*!  Þ#! #&!!!   " ! ! ! )&&)! 

  Þ"&  Þ"! "!!!!   ! " ! ! &!'#! 
 Þ!&  Þ$!
 µ 
Þ(&  Þ!&

  Þ"& Þ)! &!!!!   ! ! ! " "!$##! 

 Þ$!  Þ"! Þ*!  Þ#& $!!!! ! ! " ! *'"'!
 Þ#!  Þ"!

These calculations were carried out to many decimal places, but the displayed
numbers are rounded, at the end, to the nearest unit ($)

 )&&)! 
 &!'#! 
The equation (***) will be satisfied if Bœ , that is,
 "!$##! 
*'"'!

if W" produces 85580 units ($) ,

W# produces 50620 units ($) , ÞÞÞ etc.

Notice: the row reduced echelon form shows that M  G is invertible, so the solution for
B is actually unique: B œ ÐM  GÑ" . Þ (We could have solved the matrix equation
ÐM  GÑB œ . in the first place by trying to find M  GÑ" , but that's even more work.)

This theorem is too hard for us to prove, but we can give an intuitive argument for it.

Theorem (using the same notation as above) If G and . have nonnegative entries and
the column sums in G are all  " (that is, if every sector is profitable), then ÐM  GÑ
must be invertible and therefore there is a unique production vector satisfying the
equation ÐM  GÑB œ . , namely B œ ÐM  GÑ" . . Moreover, this B is economically
feasible in the sense that all its entries are   !Þ
Here is an informal, imprecise discussion that might give you some confidence that the
theorem is actually true.

1) Suppose G is any 8 ‚ 8 matrix with entries all   0 and all column sums  "Þ
Then G 7 Ä ! as 7 Ä ∞.
Å
(the 8 ‚ 8 zero matrixÑ

This statement just means that “each entry - in the matrix G 7 can be made as
close to ! as we like by choosing 7 large enough.”
To keep things simple, we will only look at the case when G œ 
- .
+ ,
,
a # ‚ # matrix. But the idea is similar when G is 8 ‚ 8.

Let α œ the larger of the two column sums in G . By definition of α, +  - Ÿ α

and ,  . Ÿ α Ðone of the two actually equals α.) By our assumptions above
about G , we see that ! Ÿ α  ".

An observation: suppose Bß Cß Dß A   !Þ What happens to column sums when a

# ‚ # matrix 
D A
B C
is multiplied by G on the left?

G  œ   œ
-C  .A 
B C + , B C +B  ,D +C  ,A
,
D A - . D A -B  .D

so the column sums satisfy:

Sum Column 1 œ Ð+  -ÑB  Ð,  .ÑD Ÿ αB  αD œ α ÐB  DÑ

Sum Column 2 œ Ð+  -ÑC  Ð,  .ÑA Ÿ αC  αA œ α ÐC  AÑ

In other words: for each column in the product G 

D A
B C
,

column sum in product Ÿ α Ðold column sum in 

D A
B C
Ñ

Use this fact, letting   œG œ

.
B C + ,
Þ Then
D A -

G  œ G# œ 
.  - .
B C + , + ,
and
D A -

sum for column " of G # Ÿ α (old column 1 sum in G ) œ αÐ+  -Ñ Ÿ α † α œ α#

sum for column 2 of G # Ÿ α (old column 2 sum in G ) œ αÐ,  .Ñ Ÿ α † α œ α#
.
 Repeating the calculation again using 
D A
B C
œ G # , we see that in the matrix

G$ œ G † G# œ 
.
+ ,
† G # , the sum of each column is Ÿ α$ .
-

Continuing in this way, we get that each column sum in G 7 is Ÿ α7 Þ Since all
the elements in G 7 are nonnegative, this means that every individual entry - in
the matrix G 7 satisfies ! Ÿ - Ÿ α7 Þ Since α7 Ä ! as 7 Ä ∞ (because
! Ÿ α  ")ß we get that each entry - in G 7 approaches ! as 7 Ä ∞Þ

2) M  G 7" œ ÐM  G ÑÐM  G  G #  ÞÞÞ  G 7 Ñ

Why? Just multiply it out to verify:

ÐM  GÑ ÐM  G  G #  ÞÞÞ  G 7 Ñ
œ ÐM  GÑ M  ÐMGÑ G  ÐMGÑ G #  ÞÞÞ  ÐMGÑ G 7
œ M #  GM  MG  G #  MG #  G $  ÞÞÞ  MG 7  G 7"
œ M  G  G  G #  G # ÞÞ. ÞÞÞ  G 7  G 7  G 7"
7"
œM G

3) Since G 7 Ä ! as 7 Ä ∞, it follows that for large 7 G 7" ¸ !

and therefore
M ¸ ÐM  G ÑÐM  G  G #  ÞÞÞ  G 7 Ñ

Å so ÐM  G  G #  ÞÞÞ  G 7 Ñ is
approximately an inverse for ÐM  G Ñ, that is

ÐM  G  G #  ÞÞÞ  G 7 Ñ ¸ ÐM  G Ñ"

This approximation suggests that M  G really is invertible, and, since

ÐM  G  G #  ÞÞÞ  G 7 Ñ has nonnegative entries, it also suggests that all the entries in
the exact matrix ÐM  GÑ" will be nonnegative.

4) If ÐM  G Ñ is invertible and ÐM  G Ñ" has all nonnegative elements, then of

course there is a unique solution B œ ÐM  G Ñ" . and ÐM  G Ñ" . has all
nonnegative entries.
Having done all this, we can also write:

ÐM  GÑ" ¸ ÐM  G  G #  ÞÞÞ  G 7 Ñ, so

B œ ÐM  GÑ" . ¸ ÐM  G  G #  ÞÞÞ  G 7 Ñ † . or

B ¸ .  G .  G # .  ÞÞÞ  G 7 .

Does that make any sense? Think about it like this: we ask the sectors to meet the final
demand . , and they set out to do it. But producing the goods listed in . creates
additional demands: each sector has demands from the others to be able to produce
anything at all. In fact, the sectors must supply each other with the products listed in G.
so that they can produce . . But then, to produce G . ß the sectors require an additional
GÐG . ÑÑ in products  and so on and so on .... To make it all work, the needed
production B is an infinite sum:

B œ .  G .  G # .  ÞÞÞ  G 7 .  ÞÞÞ

But because G 7 Ä ! as 7 Ä ∞, we approximate B ¸ .  G .  G # .  ÞÞÞ  G 7 .

for large 7.

The column vectors in ÐM  GÑ" actually have an economic interpretation: see the
remarks following the next example

Example Here is an example with a slightly more realistic set of data involving the
Leontief Open Economy Production Model. The data is from #13, p. 137 in the textbook.

The consumption matrix G is based on input-output data for the U.S. economy in 1958,
with data for 81 sectors grouped here (for manageability) into ( larger sectors: (1)
nonmetal household and personal products (2) final metal products (such as autos) (3)
basic metal products and mining (4) basic nonmetal products and agriculture (5) energy
(6) services and (7) entertainment and miscellaneous products. Units are in millions of
dollars. (From Wassily W. Leontief, “The Structure of the US Economy”, Scientific
American, April 1965, pp. 30-32).

0.1588 0.0064 0.0025 0.0304 0.0014 0.0083 0.1594

0.0057 0.2645 0.0436 0.0099 0.0083 0.0201 0.3413
0.0264 0.1506 0.3557 0.0139 0.0142 0.0070 0.0236
G œ 0.3299 0.0565 0.0495 0.3636 0.0204 0.0483 0.0649
0.0089 0.0081 0.0333 0.0295 0.3412 0.0237 0.0020
0.1190 0.0901 0.0996 0.1260 0.1722 0.2368 0.3369
0.0063 0.0126 0.0196 0.0098 0.0064 0.0132 0.0012

It turns out that M  G is invertible (as it should be, according to the Theorem stated)
  (%!!! 
 &'!!! 
 
 "!&!! 
 
For a final demand vector . œ  #&!!! , the solution to B œ G B  . is easy with
 
 "(&!! 
 
 &!!! 
"*'!!!

computer software such as Matlab. Rounded to 4 places, Matlab gives

1.2212 0.0271 0.0226 0.0677 0.0135 0.0227 0.2167

0.0432 1.4046 0.1244 0.0466 0.0404 0.0516 0.5103
0.0806 0.3387 1.5927 0.0555 0.0508 0.0326 0.1810
ÐM  GÑ" œ 0.6732 0.1905 0.1763 1.6448 0.0948 0.1266 0.3265
0.0636 0.0531 0.1010 0.0897 1.5393 0.0575 0.0590
0.3409 0.2711 0.2953 0.3253 0.3842 1.3674 0.6371
0.0213 0.0303 0.0392 0.0231 0.0175 0.0211 1.0246

 (%!!!   99580 
 &'!!!   97700 
   
 "!&!!   51230 
"    
so B œ ÐM  GÑ . œ ÐM  GÑ  #&!!!  œ  13157 
   
"

 "(&!!   49490 
   
 &!!!   13840 
"*'!!! 32955

(entries rounded to whole numbers)

An economic interpretation for the columns of ÐM  G Ñ"

Suppose G is an 8 ‚ 8 consumption matrix with entries   ! and column sums each less
than "Þ Let

Ї‡Ñ 
B œ production vector that satisfies a final demand .
?B œ production vector that satisfies a different final demand ?.

Ðfor the moment, just think of ?B and ?d as fancy names for two new production/final
demand vectors having no connection to Bß . Ñ

Then by (**) B œ GB  .
?B œ G ?B  ?.
Adding the equations and rearranging, we get

B  ?B œ GÐB  ?BÑ  Ð.  ?. Ñ

So production level ÐB  ?BÑ fulfils the final demand Ð.  ?. Ñ

At this point you can think of ?. as we would in calculus  as a change in the final
demand . , and ?B is the corresponding change made necessary in the production vector
B to satisfy the new final demand.

"
!
Suppose we let ?. œ  , that is, “increase the final demand from sector W"
!
ã

by 1 unit.”

Since ?B œ G ?B  ?. , we have ÐMGÑ ?B œ ?. , or

"
!
?B œ ÐM  GÑ" ?. œ ÐM  GÑ"  
!
ã

 col 1   col 2   col 8 

     
œ"† "   ! †  "   ÞÞÞ  ! †  " 
of of of

     
ÐM  GÑ ÐM  GÑ ÐM  GÑ

 col 1 
 
 " 
of
œ
 
ÐM  GÑ

"
!
So the ?B corresponding to ?. œ   is the first column of ÐM  GÑ" Þ
!
ã

In other words: a change of one unit in the final demand from sector W" requires a
change ?B in the production vector required to satisfy the new demand, and ?B œ the
first column of ÐM  GÑ" .

Similarly, you can show that a change of one unit in the demand of sector W# causes a
change in the production vector ?B œ the second column of ÐM  GÑ" à etc.
The column vectors in ÐM  GÑ" are not just “accidental” vectors that pop up in
computing the inverse; they have economic meaning.

"
!
So, in the preceding example: a change of ?. œ  ß that is a change of  1 unit
!
ã

Ð œ  " million dollarsÑ in the demand for goods from sector W" forces the necessary
production vector
 99580 
 97700 
 
 51230 
 
 13157  to change to B  ?B,
 
 49490 
Bœ
 
 13840 
32955

 1.2212 
 0.0432 
 
 0.0806 
 
?B œ "=> column of ÐM  GÑ" œ  0.6732 
 
 0.0636 
where
 
 0.0213 
0.3409